The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructions direct the agent to install the officecli utility by downloading and executing scripts directly from a remote GitHub repository using curl | bash and irm | iex. This method of installation executes unverified code from the internet and is a high-risk pattern.
[COMMAND_EXECUTION]: The skill utilizes the officecli tool to interact with the local file system and perform complex operations on .docx files, including structure modification and raw XML injection.
[PROMPT_INJECTION]: An indirect prompt injection surface is present because the skill extracts and processes text from external Word documents.
Ingestion points: Document content is read into the agent's context using the view text, view annotated, and get commands.
Boundary markers: There are no delimiters or instructions to the agent to disregard instructions that may be embedded within the document content.
Capability inventory: The skill has the ability to write to the file system and modify document internals via the officecli tool.
Sanitization: Extracted document content is processed directly without filtering or sanitization.

officecli-docx